Dear Lister,
I have collected data in 6 geographical areas on prevalence of a
parasite in humans and in foxes. The results are expressed as a number
of positive or negative cases in human and foxes in the following
data.frame:
Pvtab <-
structure(list(posHum = c(3, 5, 3, 17, 0, 4), negHum = c(32631,
16293, 27988, 231282, 53215, 51046), posFox = c(18, 23, 18, 191,
12, 55), negFox = c(14, 24, 62, 105, 55, 43)), .Names = c("posHum",
"negHum", "posFox", "negFox"), row.names = c("zone 1", "zone 2",
"zone 3", "zone 4", "zone 5", "zone 6"), class = "data.frame")
I want to check a possible link between prevalences in humans (the
reponse variable) and prevalences in foxes (the independant variable). I
though about a logistic regression of the form:
pvFox<-Pvtab$posFox/(Pvtab$posFox+Pvtab$negFox) # computes the
prevalence in foxes for each area
mod0<-mod0<-glm(cbind(Pvtab$posHum,Pvtab$negHum)~pvFox,family=binomial)
But in this cas the number of foxes that have been used to compute the
prevalence estimate in foxes (pvFox) is deliberatly not taken into
account in the model. I can hardly figure out how to do it (weighing the
model with the square root of the number of fox in each area ?).
Any advise appreciated about how to model a prevalence as a response of
another prevalence at best.
Patrick
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.